Big Data in Insurance Underwriting: Boon or Bane? Benno Keller 17 th Meeting of The Geneva Association s Annual Circle of Chief Economists Insurance Prospects in a Changing Risk Environment 24-25 March 2015, Paris www.genevaassociation.org info@genevaassociation.org
Big data in insurance underwriting: boon or bane? Annual Circle of Chief Economists Paris, March 25 Benno Keller GAIA
High expectations of big data in insurance Insurers are investing in predictive analytics 2
Concerns about big data are equally high Regulation has restricted the use of information in recent years 3
Insurance in the Garden of Eden The value of premium risk insurance 4
The benefit of knowing risk levels The value of risk mitigation and prevention Investments in risk mitigation and prevention are welfare enhancing as long as the reduction in expected loss exceeds costs for mitigation Big data is likely to reduce the cost of risk mitigation, e.g.: Targeted investments Tailored incentives Risk-based premiums provide mitigation incentives Net welfare effect of premium differentiation = welfare loss of premium risk insurance + welfare gain of improved risk mitigation 5
There is no way back to Eden The cost of adverse selection Banning the use of risk information leads to adverse selection Cost of adverse selection: Average Expected Loss low high Risk aversion high low medium low medium high 6
Bringing it all together Likelihood of cost of adverse selection outweighing benefit of premium risk insurance low high Gender in motor insurance Genetic information in life insurance Genetic information in life insurance Telematics Geo-location in property insurance low Likelihood of risk indicator affecting risk mitigation high 7
Banning the use of gender in car insurance Not an issue after all? Value of premium risk insurance Low given even distribution of risk types (men and women) across population Cost of adverse selection Low given that car insurance mandatory in many countries Relatively large maximum expected loss (3 rd party liability) Benefit of risk mitigation and prevention Not gender-related Banning the use of gender as a risk indicator for motor insurance is unlikely to significantly reduce welfare 8
Banning the use of genetic information in life insurance (protection products) Value of premium risk insurance High value of premium risk insurance as genetic predisposition may signal high probability Cost of adverse selection Currently low as most individuals do not possess genetic information As genetic testing becomes affordable and convenient, this may change and the cost of adverse selection may increase Benefit of risk mitigation and prevention For some illnesses prevention may play an important role Banning the use of genetic information in life insurance is likely to involve an increasing social cost 9
Banning the use of geo-location in property insurance Value of premium risk insurance Value of premium risk insurance may be considerable in highly exposed areas Cost of adverse selection Cost of adverse selection high for non-mandatory insurance Benefit of risk mitigation and prevention Large potential from mitigation and prevention Banning the use of geo-location is very likely to be welfare-reducing 10
Banning the use of telematics / activity trackers Value of premium risk insurance Limited value of premium risk insurance as risk is largely driven by behavior Cost of adverse selection Low cost of adverse selection Benefit of risk mitigation and prevention Potential to promote prudent behavior by policyholders Telematics / activity trackers are likely to be welfare-enhancing 11